> > I've done something similiar, but went with what I perceived to be the > somewhat less complex route of advertising the routes from both places > at > all times, but with AS Prepending and such so that Site A was always > preferred when reachable. > > It worked well, and no DNS games required for failover. >
the main problem i've seen with this is that a lot of providers are going to local pref their downstreams such that they are not using provider/peer links for traffic going to directly connected customers. given local pref beats as path length, you are likely going to get traffic bleeding over to your DR site from ISP A/B local networks.. to achieve the same goal perhaps (albeit not in the best interests of the global bgp table) advertise additional prefixes. eg primary site advertise 2 x /24 , backup site advertise 1 x /23 primary site disappears, falls through to secondary site _______________________________________________ cisco-nsp mailing list cisco-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
